By Lisa Yorgey Lester
How three direct marketers merged multiple data streams.
Data integration—the process of marrying data from disparate databases—empowers a marketer to maximize its marketing programs. A more complete picture of its customer base presents many opportunities, such as cross-selling and upselling promotions, reducing customer churn, and identifying to which channels customers best respond.
Pulling multiple data streams from stand-alone systems, merging and massaging that data, and extracting actionable information to enhance marketing strategies takes careful planning and plain old hard work. Here are the experiences of three marketers who embraced various data integration projects to address specific marketing dilemmas, and what they learned in the process.
Jenny Craig's Dirty, Scattered Data
Weight management company Jenny Craig wants to refocus its marketing efforts on its client successes rather than its products. To track a client's progress with the weight management program, it needs to rely on data pulled from more than 450 weight-loss centers.
Local centers collect all information emanating from a visit, and enter both leads and customer records in their databases. If a lead converts to a customer, then food purchases, weights and any other data collected, such as birth date and phone number, also feed into the system.
Because each center maintains its own separate legacy database, clients are married to a single center. If a client moved and started visiting another center, a new account was created. This left Jenny Craig with no way of tracking the progress of its predominately female client base.
In October 2002, Jenny Craig began to clean up its data and build a centralized data warehouse.
The company's first task was to take a good look at its data, which exposed a glaring problem with data entry. After running the data through analysis software, it discovered it had dates as early as the 1800s and as recent as 2003 entered into the birth date field. (While the company hopes to collect this information, it is not required.) It became obvious that if a client did not supply a birth date, a false date was created. This revelation exposed the need for decision rules within the system to flag improbable dates, as well as the need for better data entry training.